Hello Hack Chat!
Hi Chris - welcome!
Looks like we've got a good turnout, so let's kick it off. Welcome to the Hack Chat everyone, thanks for coming, and thanks to Chris Gammell for hosting. When Chris suggested low-level analog measurement as a topic, I jumped at it - figured we'd all love to hear about that.
Take it away, Chris!
how low level are we talking here?
obviously you all knew that
so maybe I'll step back
and mention a bit about my background
and why I thought this might be a good topic
I went to school for electronics and struggled with it
still do, some days
but then I had the opportunity to work at Keithley Instruments in Cleveland OH
and that's where I really got an education in electronics
that was working on supporting things like the 2400 source meter, 6517B electrometer, 6485 current source, etc
of course, that's been a while now since I was ther
but the low level measurement stuff sticks with you
namely working with signal processing in the analog domain
working with op amps, resistors, capacitors, etc
What signal to noise ratio?
and there are a lot of things to consider
like power cleanliness and such
and very in depth topics for test and measurement
but that was the genesis of thinking it'd make a good hack chat topic
since then, I mostly work on systems that are sensing something in the real world
but usually are piping back data over an i2c or SPI bus
so there's less direct interaction with analog signals
but most people still need to think about things in terms of analog signals and how they look to the chips they're interacting with
so I guess there are some general tips and tricks, but not sure if that's what people are interested in
or if they have specific questions they'd like me to answer
but I'm happy to do either
and can fill in as necessary
from a day-to-day basis, I think about troubleshooting and how the interaction with the analog side of things impacted me
If you had a "Top X tips and tricks" to recommend, that pertain to this subject, that might be cool!
why don't you give us an example description of a circuit you have used to buffer a signal?
so on the tips and tricks side of things
I'm always pulling out a scope first and foremost
and checking the rails
obviously the voltage levels
but also how consisten, or "clean", the power supply looks
this also usually allows me to check if I've turned the device in question on
which is a pretty common problem as well
also, if you had some solid book or other resource recommendations on the subject, maybe both historically, as well as current day
once I know the power source is looking good, I'm usually pulling out a schematic and writing down the signals and levels I expect to see, BEFORE probing the circuit
How do you juge that ? Maybe it's just me, but even measuring a _battery_ I can see noise spikes .... (coming from the environment presumably). So what's the proper way to measure a rail and make sure you see the rail itself and not ... the room.
it's important to create some mental models of what you expect to see before you actually start probing around
How much does power supply cleanliness affect readings e.g. if I'm using an opamp to buffer a signal, what should I do in respect of the opamp power rails?
I don't know about you, but my brain will do backflips to say, "That looks right-ish"
> How do you juge that ? Maybe it's just me, but even measuring a _battery_ I can see noise spikes ....
Well if it's on a battery and you see spikes, something is wrong
so in that case, I'd isolate the battery by itself
I expect to not see any spikes
but if the battery by itself does...uh...bad battery!
also, run away!
@Dustin Sackett Keith Hemingway's "Electronic Designer's Handbook" is clear and has some useful tips
Discover & share this Run Away GIF with everyone you know. GIPHY is how you search, share, discover, and create GIFs.
No, I expect it's coming from the scope or picking up LED lighting or something. But that's the issue, the environment is not noise free and I end up measuring that rather than the rail itself it seems.
> How much does power supply cleanliness affect readings e.g. if I'm using an opamp to buffer a signal, what should I do in respect of the opamp power rails?
Could be quite a bit! If you have a bunch of AC noise that gets rectified, it'll look like the DC signal is higher than it actually is
@Kelly Heaton , cheap and looks exhaustive! I will check that outthank you
> No, I expect it's coming from the scope or picking up LED lighting or something. But that's the issue, the environment is not noise free and I end up measuring that rather than the rail itself it seems.
Sure, the environment can really impact things
but it's all about isolation and mental models
what do you expect to see -> What do you actually see -> Why is that happening -> can you isolate it
@duddy it's much easier to understand than the Bible (Art of Electronics by Horrowitz and Hill, which you need to own anyway even just to lift weights)
in terms of isolating different parts of the circuit, I'm usually designing in 0 ohm resistor severywhere
so I can swap in a low value resistor to look at current
or to pull it entirely and isolate that part of the circuit
Is it enough to take power from an SMPS, run it through a linear regulator and put a decoupling cap on there? Or is there more to do?
I was tutoring someone recently and the 0 ohm resistor thing was a new thing to them
@Chris Gammell could you give a general recommendation on the design process for a high end low-level analog circuits?
@Chris Gammell 0 ohm resistor?
and of course, we all start somewhere! That's a huge part of my design process now, knowing how to unhook things from the rest of the ciricuit
0 ohm resistor = SMD jumper ;)
> Is it enough to take power from an SMPS, run it through a linear regulator and put a decoupling cap on there? Or is there more to do?
Depends on your application
sometimes that's all you have available
> 0 ohm resistor = SMD jumper ;)
And that's right! they're called a few different things
but adding in different "options" in a circuit can really help
but jumpers are sometimes designed into the PCB
I want to clarify that's not what I normally do
@Kelly Heaton have that Bible you speak of! I never heard about it throughout my education, and saw it recommended on here! :)
I'll design a 0 ohm resistor into the signal path
and then it's an option when i'm troubleshooting
> could you give a general recommendation on the design process for a high end low-level analog circuits?
@Dustin Sackett that's a pretty broad ask :-D
that's another thing to note about analog
most people don't do it because it's expensive as heck
like watch Shahriar or Dave Jones or anyone opening test equipment
and you'll see things like...a $5 resistor!
well, maybe just a workflow summary
@Chris Gammell I do it because its dirt cheap!
I used to work with a 1 TOhm resistor at Keithely
it was bananas
and it was really accurate too
but usually you'll end up paying for good components
Geeze, a humid room is less than 1Tohm! Hahaha
so I suppose another thing is understanding the differences between accuracy, precision and repeatability.
Napkin sketch>Schematic>Simulation>breadboard (? mixed reviews on this)>PCB layout?
Excellent points Chris, "Look for clean", "Check the rails", "Isolate using 0 ohm". Keep going!
so let's look at a simple circuit
that's a cool diagram!
like a simple circuit, right?
it's basically measuring the current coming out of (going into) a photodiode
but in the three things in that circuit, there's not much you can control
so if you want a truly accurate circuit